The A.I. Enthusiast Newsletter #11

Hello and welcome to Edition #11 of the?A.I. Enthusiast Newsletter!?

I must admit that I have read a lot of great news this week, but the one that truly grabbed my attention was the announcement made by DeepMind that their AlphaFold 2 model could predict the structures of almost all proteins known to science. Simply great, this. To better understand specific proteins, what they do, and how they work, other scientists will use DeepMind's 200x increase in the size of its protein database, which now contains over 200 million predicted structures. This will speed up scientific research and discovery worldwide and enable more breakthroughs.

I don't need to elaborate on how significant it is to the scientific community because AlphaFold has been utilized to speed up research on important real-world issues like plastic pollution and antibiotic resistance by more than 500,000 researchers. Imagine the magnitude of the global impact when the second iteration of the model doubles the size of this knowledge base.

In keeping with my excitement for science, I'm also sharing a ScitechDaily piece that describes how a new Columbia University AI program examined physical phenomena and discovered pertinent variables. To examine physical processes using a video camera and then attempt to identify the smallest possible collection of fundamental variables that accurately capture the dynamics being watched, the researchers developed an AI solution.

The researchers also claim that this kind of AI can aid in discovering intricate phenomena in fields ranging from cosmology to biology, where theoretical understanding is not keeping up with the flood of data.

This week's final story I'd like to offer is about an intriguing development demonstrating how some businesses are moving their machine learning data and models from the cloud to their in-house-managed servers. They assert that spending less money and achieving superior results are possible. You should read this if you're running any large Harry ML workloads.

And for the "Paper of the Week," I chose YOLOv7. This week, out of all the papers I had the chance to read, the brand-new, cutting-edge real-time object detector particularly piqued my interest.

YOLOv7?uses optimized modules and optimization techniques to increase accuracy at training time and reduce the params and computation of the SoTA real-time object detector by about 40% and 50%, respectively.

The authors claim that?YOLOv7?outperforms all existing object detectors in speed and accuracy in the range of 5 to 160 frames per second and has the greatest accuracy of 56.8% AP among all current real-time object detectors with 30 frames per second or higher on GPU V100.

And finally, for the "Video of the Week," I have selected a very interesting one for everyone who loves your pets; who doesn't, right? The video explains how a California-based organization wants to harness the power of machine learning to decode communication across the entire animal kingdom. Molly Gambhir tells you if we could talk to animals in the future. Check it out!

This newsletter has been moved to my blog on Medium.com. You can read it by clicking?here.

Enjoy it, and don't forget to?subscribe to the A.I. Enthusiast newsletter and share?it with your colleagues and friends.?

If you want to remain updated about the topics, I regularly write about (A.I., Business, Career Management, the future of work, and others, follow me on?Linkedin,?Twitter,?Medium,?and?Vocal.

About the author:

Jair Ribeiro, AI Strategist and Evangelist, author of "A.I.?2021: Another great year writing about Artificial Intelligence",?A.I., Robotics and Coding (for Parents): A practical guide for analog parents with digital kids,?The Terminator paradox: How neuroscience can help us to understand Empathy and the fear of Artificial Intelligence?and other books about A.I., presents his Newsletter with the most important artificial intelligence and machine learning news and articles of the week.